The properties of the web application manager differ when the web application is clustered.

Here are snippets from the PS's DebugServlet output to illustrate the situation:

A non clustered web app:
ManagedDeployment: //localhost/juddi; state: STARTED
+++ ManagedComponent(name=jboss.web:host=localhost,path=/juddi,type=Manager, type=(ComponentType

{type=MBean, subtype=WebApplicationManager}), compName=jboss.web:host=localhost,path=/juddi,type=Manager, attachment: null
++++++ properties: [sessionMaxAliveTime, duplicates, maxInactiveInterval, entropy, activeSessions, sessionCounter, sessionAverageAliveTime, processingTime, sessionIdLength, modelerType, processExpiresFrequency, algorithm, maxActive, distributable, maxActiveSessions, randomFile, name, rejectedSessions, expiredSessions, pathname, className]

A clustered web app:
ManagedDeployment: //localhost/test-simple; state: STARTED
+++ ManagedComponent(name=jboss.web:host=localhost,path=/test-simple,type=Manager, type=(ComponentType{type=MBean, subtype=WebApplicationManager}

), compName=jboss.web:host=localhost,path=/test-simple,type=Manager, attachment: null
++++++ properties: [MaxActive, RejectedSessionCount, ExpiredSessionCount, SessionIdLength, SnapshotInterval, TimeSinceLastReset, ReplicationTrigger, MaxPassivatedSessionCount, SnapshotMode, ExpiredSessions, Distributable, ProcessingTime, PassivationEnabled, MaxInactiveInterval, UseJK, ClassName, LocalActiveSessionCount, PassivationMinIdleTime, ReplicationGranularity, MaxActiveSessionCount, MaxLocalActiveSessionCount, ReplicationStatistics, SessionAverageAliveTime, MaxActiveSessions, ActiveSessions, Algorithm, RejectedSessions, PassivatedSessionCount, Name, PassivationMaxIdleTime, CacheConfigName, SessionCounter, MaxUnreplicatedInterval, ActiveSessionCount, Duplicates, SessionMaxAliveTime, MaxActiveAllowed, CreatedSessionCount, ProcessExpiresFrequency]

Notice that the similar properties like "activeSessions" and "ActiveSessions" differ only in the first letter case. These attributes seemingly mean the same thing and therefore there should be no reason why they should be called different names in clustered and non-clustered scenarios.

Note that this is causing confusion in tools like JBoss ON that can get caught by surprise by these subtle and mostly undocumented differences.